COUNT(DISTINCT field) now honors charsets

sql/item_sum.cc

@@ -542,7 +542,7 @@ void Item_sum_hybrid::reset_field()
   if (hybrid_type == STRING_RESULT)
   {
     char buff[MAX_FIELD_WIDTH];
-    String tmp(buff,sizeof(buff),default_charset_info),*res;
+    String tmp(buff,sizeof(buff),result_field->charset()),*res;
 
     res=args[0]->val_str(&tmp);
     if (args[0]->null_value)
@@ -897,17 +897,17 @@ String *Item_variance_field::val_str(String *str)
 
 #include "sql_select.h"
 
-static int simple_raw_key_cmp(void* arg, byte* key1, byte* key2)
+int simple_raw_key_cmp(void* arg, byte* key1, byte* key2)
 {
   return memcmp(key1, key2, *(uint*) arg);
 }
 
-static int simple_str_key_cmp(void* arg, byte* key1, byte* key2)
+int simple_str_key_cmp(void* arg, byte* key1, byte* key2)
 {
-  /* BAR TODO: remove default_charset_info */
-  return my_strnncoll(default_charset_info,
-		    (const uchar*) key1, *(uint*) arg,
-		    (const uchar*) key2, *(uint*) arg);
+  Item_sum_count_distinct* item = (Item_sum_count_distinct*)arg;
+  CHARSET_INFO *cs=item->key_charset;
+  uint len=item->key_length;
+  return my_strnncoll(cs, (const uchar*) key1, len, (const uchar*) key2, len);
 }
 
 /*
@@ -1036,15 +1036,23 @@ bool Item_sum_count_distinct::setup(THD *thd)
       */
       Field* field = table->field[0];
       switch(field->type())
+      {
+      case FIELD_TYPE_STRING:
+      case FIELD_TYPE_VAR_STRING:
+	if (field->binary())
+	{
+	  compare_key = (qsort_cmp2)simple_raw_key_cmp;
+	  cmp_arg = (void*) &key_length;
+	}
+	else
 	{
 	  /*
 	    If we have a string, we must take care of charsets and case
 	    sensitivity
 	  */
-      case FIELD_TYPE_STRING:
-      case FIELD_TYPE_VAR_STRING:
-	compare_key = (qsort_cmp2)(field->binary() ? simple_raw_key_cmp:
-				   simple_str_key_cmp);
+	  compare_key = (qsort_cmp2)simple_str_key_cmp;
+	  cmp_arg = (void*) this;
+	}
 	break;
       default:
 	/*
@@ -1052,10 +1060,11 @@ bool Item_sum_count_distinct::setup(THD *thd)
 	  be compared with memcmp
 	*/
 	compare_key = (qsort_cmp2)simple_raw_key_cmp;
+	cmp_arg = (void*) &key_length;
 	break;
       }
+      key_charset = field->charset();
       key_length  = field->pack_length();
-      cmp_arg = (void*) &key_length;
       rec_offset  = 1;
     }
     else // too bad, cannot cheat - there is more than one field

sql/item_sum.h

@@ -156,6 +156,7 @@ class Item_sum_count_distinct :public Item_sum_int
   TMP_TABLE_PARAM *tmp_table_param;
   TREE tree;
   uint key_length;
+  CHARSET_INFO *key_charset;
   
   // calculated based on max_heap_table_size. If reached,
   // walk the tree and dump it into MyISAM table
@@ -175,6 +176,8 @@ class Item_sum_count_distinct :public Item_sum_int
   int tree_to_myisam();
 
   friend int composite_key_cmp(void* arg, byte* key1, byte* key2);
+  friend int simple_str_key_cmp(void* arg, byte* key1, byte* key2);
+  friend int simple_raw_key_cmp(void* arg, byte* key1, byte* key2);
   friend int dump_leaf(byte* key, uint32 count __attribute__((unused)),
 		       Item_sum_count_distinct* item);